On Fri, 23 May 2003 12:48:17 +0200 "Hans J. Baumgartner"
<[email protected]> wrote:
> I have a question concerning predicition after performing a heckman
> estimation :
> Do I have to calculate and add the mills ratio after having predicted
> (after using "heckman" or does the stata command "predict, xb" (after
> "heckman") automatically adjust for that, so that my results are unbiased?
What you do depends, I think, on what the aim of running -heckman- is.
Suppose you have a model for the wages of all women
log(w) = b'X + u
but w is observed only for working women.
The real parameters of interest are the "b", and they concern the
population of all women.
The -heckman- procedure is a way of estimating the "b" taking account
of the censoring /in the data/. Thus when producing estimates of
expected log(w) for the population of woman, you would predict using
b_hat'X. The mills ratio is /not/used for the prediction.
On the other hand, if you want to derive expected log wage /conditional
on participation/, then mills ratio is relevant.
There are many textbook expositions of selection models, and the
different types of prediction formulae. One of my current favourites is
chapter 7 of M Verbeek, A Guide to Moden Econometrics, Wiley, 2000.
Stephen
----------------------
Professor Stephen P. Jenkins <[email protected]>
Institute for Social and Economic Research (ISER)
University of Essex, Colchester, CO4 3SQ, UK
Tel: +44 (0)1206 873374. Fax: +44 (0)1206 873151.
http://www.iser.essex.ac.uk
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
